Advancing retinal OCT image analysis as a biomarker for Parkinson’s disease

Resumen

The number of people worldwide affected by Parkinson’s disease (PD) is increasing every year. This is a worrying trend, which requires new biomarkers to better diagnose and monitor the disease. Research to date has demonstrated that, in addition to brain neurodegeneration, there is a dysfunction of the retina in PD patients. Promisingly, several studies have shown that retinal changes in PD can be detected using optical coherence tomography (OCT) imaging. However, there is no conclusive agreement on the potential of OCT as a reliable biomarker for PD. As a key limitation, most research has focused on a small set of features and standard OCT image analysis. Applying more advanced methods could help identify the specific aspects of the retina affected in PD and potentially uncover new biomarkers. Given this context, the present body of work has two main objectives: 1) to improve and advance existing OCT image analysis techniques and 2) to apply the developed methods to be able to study the retina of PD patients in more detail. First, novel methods were developed for OCT image alignment, quality control, and feature extraction. The developed methods were integrated into an open-source toolbox called RETIMAT, which forms the backbone of this work and is freely available to the community. Subsequently, data from healthy controls was utilized to evaluate the impact of age and sex on retinal morphology and create a normative dataset. The potential of OCT features for the diagnosis, severity assessment, and monitoring of PD was then evaluated, employing both conventional and novel features. Data for this purpose was obtained from two of the largest longitudinal cohorts to date. Our findings reveal that the explored OCT features contain information related to cognitive and motor impairment. Furthermore, the evidence suggests a differential evolution of the retina of PD patients over time. However, accurate diagnosis, severity assessment, and monitoring at an individual level using OCT features does not appear feasible. In can thus be concluded that OCT may be a better tool for understanding general aspects of PD than for guiding clinical decisions on individual patients.